Interpretable Context Methodology

Build your AI workspace in minutes, not hours

Answer a few guided questions. Get a fully structured folder system that tells Claude where it is, what to do, and where to put the work — every session, from day one.

Already have a workspace? Re-open it to edit and re-export.

Drag your workspace-state.json here, or click to browse

How it works

1

Pick your archetype

Content creator, freelancer, developer, small business — or define your own.

2

Answer guided questions

Each question explains the ICM principle behind it. No guessing why it matters.

3

See the architecture

An interactive diagram shows how all your files connect. Click any node to preview it.

4

Download and use

Get a .zip with every file fully populated — not blank templates. Open in Claude Code, Cursor, or any editor.

The 6-layer ICM system

L0

CLAUDE.md — Always loaded. The map. Identity, routing table, naming conventions. ~800 tokens.

L1

CONTEXT.md — Loaded on workspace entry. Stage map, how stages connect, routing logic.

L2

Stage contracts — One per stage. Loaded per task. Inputs / Process / Output / Done looks like.

L3

_config/ files — Loaded selectively. Voice, format patterns, constraints, business rules. Stable across runs.

L4

output/ directories — Working files. Stage N's output is Stage N+1's input. Human review at each handoff.

L5

skill-starters/ — Optional prompt templates. Reusable “skills” to start work in each stage quickly and consistently.

◇ Build with AI conversation Describe your project — Claude will pre-fill the wizard for you

Workspace:

Generated Files

← Click any file to preview its content